Building fully flexible tail (several joints) with minimal actuation required

I’m looking to build something similar to the tail of an animal. I would like to have as many joints as possible (think piece of roller chain turned up on it’s side) to get a semi free-flowing motion out of it. Here’s the catch: I can only use one or two hydraulic cylinders (and any assortment of linkages or cables) to actuate it. If possible, I’d like to get a free flowing motion similar to a snake.

For now, let’s say this thing only has to move in the horizontal plane, and only needs to be able to go from straight, to curved in one direction. Let’s say it does not need to swing the other way.

I’d just trying to get conceptual ideas if there is any way to do this. If knowing the size helps, hold on to your armrests. I’m looking at a total of approximately 8 feet long give or take a couple.

Any ideas are welcome, as long as they are explained so I can understand. :slight_smile: Thanks in advance.

A little study of anatomy will show you that the motion you need can be achieved with cables on both sides of your device. Pull on the cable on one side and the device will naturally curl in that direction. Of course you need to secure the cable at each of the joints but that is easy to figure out. A few springs will allow the “tail” to straighten when the tension is released.

Geez, you read a post, decide to reply later, and got to sleep for 8 hours, then Al takes your idea…

Anyway, consider a chain, like roller chain that would go on a sprocket. Put it on a table with the pins perpendicular to the surface of the table (ie if it were on a sprocket and the sprocket was lying flat on the table). With the chain straight, run a cable along the side. Put some sort of guide on each link and attach the cable to the upper link. If you pull on the cable with your actuator, it’ll curl in the direction of the cable. Now, run a cable on the other side of the chain and keep your “tail” curled in this direction using a spring. If you were to completely relax the actuator on the first side, the “tail” would naturally curl toward the spring side. You can then use the actuator to straighten it by pulling the cable on its own side. You can now make it curl in both directions using only one actuator. This however, would only get you one big curl, not the snake-like motion you wanted.

You only asked how to curl it in one plane, but I wanted it to go in the perpendicular plane also. To make it curl perpendicular to the table’s surface, you’d need the other actuator (which is why I only used one above). Obviously, you’d also need to find a replacement for the chain because it doesn’t bend in that direction. If you put the same cable setup in the perpendicular plane, you now have 3-D range of motion.

Us old guys don’t need as much sleep.

Thanks for all the ideas. I think one big smooth curl is okay. I’m going to experiment on a smaller scale as soon as I can. :slight_smile:

Actually you wouldn’t attach it to each joint. You have something like described above except drill holes through each piece and connect the cables to the end piece. Ive seen a snake toy similiar to this. It completely freaked people out because its movements were 100% real looking!

what if you used springs?

use what you can for hydralics, and then use the springs on the joints to keep it realistic and moving. They should be easy enough to work with minimal actuation.

Mechanicalbrain, I think I actually had a toy snake sort of like that at one time or another. But it wasn’t actuated, you just held it out and tilted it a little and gravity would make it move. It was a lot like plastic roller chain actually.

Anyway, I tried to make a model this thing with a piece of roller chain (#35) and a rubber band (#33) cut so it would be straight and not in a loop. I tied it to two points on the chain such that when the chain was straight, the rubber band would be under tension. I then tied a piece of string to the end of the chain, and ran it down the other side. I then placed a small zip tie at every link to keep the string and the rubber band close to the chain. Anyway, the results were quite marginal at best. Some links would want to move and some would want to stay. The biggest trend I noticed is that some links would want to move in the opposite direction from the direction the chain was moving as a whole. This resulted in the chain pivoting at only one or two points out of maybe 12. Basically, the chain was attempting to stack on itself every time the string was pulled. For the tail to properly curl (in one direction) I would need to have stops on each “link” so that it could not pivot in what we’ll call the negative direction (the other way past being straight). And even then, I’m still not sure it would work, but I think there’d be much better chances.

What I want it to be able to do is go from being straight, to being sort of like this but on it’s side.

And now here’s the really hard part. This thing’s motion has to look good on TV (slow and smooth) and it will be all covered in flowers. There mustn’t be anything visible except for flowers (I believe carnations to be exact) so it will need some sort of flexible skin/covering.

I cannot tell a lie, I saw this first demonstrated for the tail on Jabba the Hut in Star Wars. Just the right tension and balance on a few springs and a mechanical joystick was able to perform some pretty good motion.

To do more than a small curve, you need the wires you are pulling to be able to extend into the gap between the curled up part to the end. Reguarding what Al said, Jabba’s tail had covering enough and flab enough so that going from the tip back to the point they were pulling wasn’t hard. It was already covered. See the pic below for what I am trying to say. If you try to keep the wire (small line) next to the tail (big line), you will be pulling agenst youself after a little while.



why dont you make it overly complicated and use a bunch of servos? but i assume its going to be bigger that what im picturing it, but a bunch of standard RC servos would work really well, you could also get a board too, and make all the servos sync up with eachother

Are you building a parade float?

Not really. The wires were attached at each joint with a pulley system so that there was no need for wires “out in the air”. When you pull back on the wire which is attached to the tip, it pulls on the tip until the joint runs out of mechanical play (hits a premade mechanical stop.) Then the next joint flexes until the same thing happens. You don’t get the exact motion of a true muscle structure since you are only really using one “muscle” but the approximation is pretty good. Similar mechanics were used in the original “Jaws” but had fewer joints and needs far more power.

I would think that attaching it to every joint would limit the movement and the amount of curve becuase you would start fighting youself, especially if you what it to curve back like in sanddrag’s picture. But I might have a serious misunderstanding of what I’m talking about. It makes sence in my own head.

You may be envisioning the chain which has nearly +/- 180 dregrees of motion at each joint. Think of something that only has +/- 10 degrees. Monkees with prehensile tails can have motion that allows the tail to wrap around a limb, but to duplicate that with mechanical/electrical devices is a challenge. NASA has been working on this long and hard. A friend of mine was experimenting with “muscular motion” by stretching wire across a bladder. Inflate the bladder and the wires are pulled taut.

Is that something like the no-rod pneumatics thing discussed here? (

grr, internet explorer doesn’t work well with this interface…

I believe what I’m attaching is an illustration of the idea Al first mentioned, though using rotary rather than linear actuation. There are other methods of course, for implementing both linear and rotary actuation, including employing a simple cam.

It produces motion in a single plane and can curve to both the left and right – though not both simultaneously.



First off the cable wouldn’t be attached to each joint. Put a hole on each side of each segment and you only connect the cable to both sides of the last segment.

The cable is not attached to any segment but the first and last, it is simply captured by pulleys to keep it close to each segment when in tension or relaxed.

I was responding to Al’s post. It doesnt need all those pulleys. You know i keep thinking of those toy robot-like hands that you wore over your own. Each one of your fingers go in a ring and the finger was extended. Ill post a picture.